Five studies were included (n=48, range six to 19): one controlled study with matched pairs (n=8); and four before-and-after studies (n=40), one of which had a relevant control group. Study quality was low (median PEDro score 4 points, range 3 to 6). None of the studies used allocation concealment and only one used blinded outcomes assessment.
In two studies the intervention improved self-selected walking speed over 10 metres, with large effect sizes; however, the effect was statistically significant in only one study (1.48, 95% CI 0.49 to 2.4; n=14). A third study reported a significant increase in walking speed over 10 metres. A fourth study reported a large but non-significant effect from the intervention among more severely affected children (effect sizes not calculable). Studies reported no statistically significant effect on walking endurance (two studies) and physical support required to achieve functional ambulation (two studies).
All four relevant studies reported an improvement from the intervention in Dimension E of the Gross Motor Function Measure, with a large effect in two studies; however, the effect was statistically significant in only one case (1.52, 95% CI 0.52 to 2.51; n=10). Two of three relevant studies reported an improvement in Dimension D, one with a large effect size (1.00, 95% CI 0.07 to 1.93; n=10). One study reported a significant improvement in overall Gross Motor Function Measure score, which was retained for at least 12 weeks post-intervention.
One study reported that the intervention improved energy expenditure, with a large effect of borderline statistical significance. A second study reported no significant improvement in muscle tone or selective motor control.
All studies reported that training was well tolerated by participants.